I have been using Google Home Mini, but its AI seems stupid. It annoyingly lacks customisability so I have make some unnecessary efforts to work around it. Also, it has a dumb bug that, even though I have reported it to Google last year, has not been fixed. So, I want to ditch it and create my own for my own use. I have found that the features I need are very limited, and they do not really need any A.I to process them. All I need is (1)to catch my speech and (2)convert it into text (offline, of possible). After that, I can easily write code to react to them.
For (2), a quick search gave me something called "deepspeech". So, I wonder what hardware I should buy for (1). I have searched the web and many search results were using "ReSpeaker". Of course, I am not rich, so I do not want to spend a lot of money, if possible. Is "ReSpeaker 2" the best hardware for my case? I would like to have Google-Home-Mini-like catch range. That is, being able to speak to my Raspberry Pi 4 2 or 3 metres away. There are some search results that are using regular USB microphones, but I am not sure if they could get the speech 2~3 metres away.